(好一些的综述)https://arxiv.org/abs/2506.18096
(过于求全反而缺乏洞见的综述) https://arxiv.org/pdf/2506.12594
(评估方法,从报告质量和引用精度两个角度评估)https://arxiv.org/pdf/2506.11763
(推荐,带UI,字节开源的基于LangGraph的DeepFlow)https://github.com/bytedance/deer-flow…
(推荐,带UI,Gemini基于LangGraph的Deep Search实现,比较简单) https://github.com/google-gemini/gemini-fullstack-langgraph-quickstart…
(没有UI,但是功能比较丰富)https://github.com/foreveryh/mentis/tree/main/super_agents/deep_research…
(基于smolagents框架,比较简单)https://github.com/huggingface/smolagents/tree/main/examples/open_deep_research… (基于OpenAI Agents 库实现,比较简单) https://huggingface.co/spaces/mallocode200/Deep_Research_Assistant/blob/main/research_manager.py